Artificial Intelligence Policy

Transparency and human oversight of AI use

This policy covers generative, predictive, and analytical artificial intelligence used in design, searching, collection, material creation, programming, analysis, visualization, translation, writing, review, and production. It applies to public services and tools embedded in academic software.

Authors disclose and validate; reviewers protect confidential content; editors assess transparency and risk; and technical staff use only authorized functions. No automated system replaces human accountability or scientific decision making.

Permitted use

Artificial intelligence may support translation, language editing, programming, exploratory analysis, information search and organization, or auxiliary materials under human supervision.

Mandatory disclosure

Authors must identify the tool, version or date when relevant, purpose, section or stage, scope of intervention, and human verification procedure.

Limits

Artificial intelligence cannot be an author.

It cannot fabricate data, findings, or references.

It cannot conceal plagiarism or deceptively manipulate images.

Manuscripts, personal data, or confidential information must not be uploaded to unauthorized services.

Authors retain full responsibility for methods, citations, results, and conclusions. Reviewers must not use generative AI to produce substantive reports or compromise confidentiality.

Use in research

When an artificial intelligence tool forms part of the method—for example, classification, adaptive tutoring, educational content generation, or automated analysis—the procedure must be described in enough detail for assessment and replication. Authors report the name and version, provider, access date, input data, relevant settings, validation, human oversight, performance metrics, known biases, and personal-data handling.

AI outputs are not accepted as evidence by themselves. Authors must verify accuracy, document errors, and compare performance through appropriate procedures. Educational studies should examine, where relevant, effects on learning, equity, privacy, accessibility, and teacher and student autonomy.

Use in writing and presentation

Limited assistance with clarity, grammar, or translation should be disclosed when it materially changes the text. Generation of sections, images, code, analyses, or references requires a specific manuscript disclosure. Outputs must be checked against original sources; nonexistent references and unsupported claims are unacceptable.

An AI tool cannot meet authorship requirements because it cannot approve the final version, disclose conflicts, or assume accountability. Human authors remain responsible for originality, legality, accuracy, and the absence of infringement in all assisted content.

Editorial use

Editors and reviewers must not upload manuscripts or unpublished data to external systems that may store, reuse, or train on the content. Authorized internal functions may be used only with confidentiality, security, and human oversight controls. AI may not independently determine acceptance or rejection.

Concealed substantive use, AI-assisted fabrication of data or references, and deceptive image manipulation are investigated as potential breaches. This policy is reviewed periodically as technology and regulation evolve, while human accountability remains mandatory.

AI disclosure and verification

Identify the tool, provider, version or access date, and specific purpose.

Describe inputs, relevant settings, outputs used, and validation when AI forms part of the method.

Disclose manuscript sections, images, code, analyses, or translations materially assisted by AI.

Check references, facts, calculations, bias, originality, licenses, and possible personal data.

Exclude tools from authorship and keep human authors accountable.

Do not upload manuscripts, reports, or sensitive data to unauthorized external services.

Retain prompts, settings, code, and validation records when needed for reproducibility.

Documentation of AI-assisted processes

A statement of use or absence of substantive AI use.

Sufficient methodological description when AI is the object or instrument of study.

Documentation of human validation and error correction.

Editorial authorization for any internal use processing confidential content.

Editorial verification of AI declarations

Where AI is declared for the analysis of educational evidence, the handling editor shall request information sufficient to assess reproducibility: purpose, inputs, validation procedures, limitations, detected errors, and human oversight. Missing disclosures shall trigger clarification; suspected fabricated citations or data shall be handled under the integrity policy, without treating an automated detector as conclusive proof.

Updates to AI criteria

The policy is reviewed at least annually as technology, regulation, and educational use evolve. The journal may request additional information or validation materials. Deliberate omission, invented references, fabricated data, or deceptive manipulation is handled under the ethics policy.

Artificial intelligence as research subject and research tool

Where AI is the subject of research—for example generative tutors, dropout prediction, learning analytics or automated assessment—authors should report model provenance or training-data information where available, version, date, setting, metrics, comparators, human involvement, bias and student risks. A technical demonstration without an educational research question, evaluation design or critical analysis is insufficient.

When AI assists research or writing, editors must distinguish language-level assistance from uses affecting analysis, qualitative coding, instruments, inclusion decisions or synthesis. Confidential OJS content must not be entered into external generative services without documented safeguards and applicable authorization.

Disclosure proportionate to use

Where AI was used solely for language assistance, the statement identifies the tool and type of support. Use in coding, classification, statistical analysis, material generation, software development or interpretation requires relevant inputs, procedures, versions where pertinent and human checks sufficient to assess reproducibility. References, quotations and factual claims must be verified against their original sources; AI cannot be named as an author.

AI in confidential assessment

Reviewers must not upload unpublished manuscripts, datasets, figures or review reports to external AI services without explicit journal authorization and adequate information-handling safeguards. Any permitted use is disclosed to the editor. Editorial recommendations remain human scholarly judgments grounded in assessable evidence.